Eduardo Costa Da Silva

Prof. Dr. Eduardo Costa Da Silva

Electrical Engineering, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil
orcid

Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 July 2019 in Sensors
Verified
Journal Article
Open access
PDF file
ISIam Palatnik de Sousa
Iam Palatnik de Sousa
Marley Maria Bernardes Rebuzzi VellascoMarley Maria Bernardes Rebuzzi Vellasco
Marley Maria Bernardes Rebuzzi Vellasco
Eduardo Costa Da SilvaEduardo Costa Da Silva
Eduardo Costa Da Silva
An application of explainable artificial intelligence on medical data is presented. There is an increasing demand in machine learning literature for such explainable models in health-related applications. This work aims to generate explanations on how a Convolutional Neural Network (CNN) detects tumor tissue in patches extracted from histology whole slide images. This is achieved using the “locally-interpretable model-agnostic explanations” methodology. Two publicly-available convolutional neural networks trained on the Patch Camelyon Benchmark are analyzed. Three common segmentation algorithms are compared for superpixel generation, and a fourth simpler parameter-free segmentation algorithm is proposed. The main characteristics of the explanations are discussed, as well as the key patterns identified in true positive predictions. The results are compared to medical annotations and literature and suggest that the CNN predictions follow at least some aspects of human expert knowledge.
1

23 August 2021 in Sensors
Verified
Journal Article
Open access
PDF file
ISIam Palatnik de Sousa
Iam Palatnik de Sousa
Marley Maria Bernardes Rebuzzi VellascoMarley Maria Bernardes Rebuzzi Vellasco
Marley Maria Bernardes Rebuzzi Vellasco
Eduardo Costa Da SilvaEduardo Costa Da Silva
Eduardo Costa Da Silva
Problem: An application of Explainable Artificial Intelligence Methods for COVID CT-Scan classifiers is presented. Motivation: It is possible that classifiers are using spurious artifacts in dataset images to achieve high performances, and such explainable techniques can help identify this issue. Aim: For this purpose, several approaches were used in tandem, in order to create a complete overview of the classificatios. Methodology: The techniques used included GradCAM, LIME, RISE, Squaregrid, and direct Gradient approaches (Vanilla, Smooth, Integrated). Main results: Among the deep neural networks architectures evaluated for this image classification task, VGG16 was shown to be most affected by biases towards spurious artifacts, while DenseNet was notably more robust against them. Further impacts: Results further show that small differences in validation accuracies can cause drastic changes in explanation heatmaps for DenseNet architectures, indicating that small changes in validation accuracy may have large impacts on the biases learned by the networks. Notably, it is important to notice that the strong performance metrics achieved by all these networks (Accuracy, F1 score, AUC all in the 80 to 90% range) could give users the erroneous impression that there is no bias. However, the analysis of the explanation heatmaps highlights the bias.

15 February 2011 in Measurement Science and Technology
Verified
Journal Article
Closed access
Eduardo Costa Da SilvaEduardo Costa Da Silva
Eduardo Costa Da Silva
LGL A P Gusmão
L A P Gusmão
CHCARLOS HALL
CARLOS HALL
EME Costa Monteiro
E Costa Monteiro
FMF L A Machado
F L A Machado

14 February 2018 in Measurement Science and Technology
Verified
Journal Article
Closed access
PDF file
LBL S Benavides
L S Benavides
Eduardo Costa Da SilvaEduardo Costa Da Silva
Eduardo Costa Da Silva
EMElisabeth Costa Monteiro
Elisabeth Costa Monteiro
CHCARLOS HALL
CARLOS HALL
This paper presents a new configuration for a GMI pressure transducer based on the reading of the phase characteristics of GMI sensor, intended for biomedical applications. The development process of this new class of magnetic field transducers is discussed, beginning with the definition of the ideal conditioning of the GMI sensor elements (dc level and frequency of the excitation current and sample length) and continuing with computational simulations of the full electronic circuit performed using the experimental data obtained from measured GMI curves, and have shown that the improvement in the sensitivity of GMI magnetometers is larger when phase-based transducers are used instead of magnitude-based transducers. Parameters of interest of the developed prototype are thoroughly analyzed, such as: sensitivity, linearity and frequency response. Also, the spectral noise density of the developed pressure transducer is evaluated and its resolution in the passband is estimated. A low-cost GMI pressure transducer was developed, presenting high resolution, high sensitivity and a frequency bandwidth compatible to the desired biomedical applications.

1 April 2022 in Neural Networks
Verified
Journal Article
Closed access
ISIam Palatnik de Sousa
Iam Palatnik de Sousa
Marley Maria Bernardes Rebuzzi VellascoMarley Maria Bernardes Rebuzzi Vellasco
Marley Maria Bernardes Rebuzzi Vellasco
Eduardo Costa Da SilvaEduardo Costa Da Silva
Eduardo Costa Da Silva

1 March 2013 in Electronics Letters
Verified
Journal Article
Open access
PDF file
Eduardo Costa Da SilvaEduardo Costa Da Silva
Eduardo Costa Da Silva
LGL.A.P. Gusmão
L.A.P. Gusmão
CHCARLOS HALL
CARLOS HALL
EME.C. Monteiro
E.C. Monteiro
Over the last decade, giant magnetoimpedance (GMI) sensors have been intensively studied as a new possibility for measuring ultra‐weak magnetic fields. Presented is a newly developed method that allows the enhancement of the impedance phase sensitivity of GMI sensors and, consequently, increases the sensitivity of phase‐based GMI magnetometers. The electronic circuit that implements the proposed method is presented and discussed, and the obtained experimental results indicating its effectiveness are shown.